Laura H

Laura H said:

How do I combine two monolingual yaml files (one for source, one for target language, similar to xml) in Studio for a partially existing translation project?

Is this to create a TM with the already translated segments?  Assuming it is then just align the files, or at least the segments within them, to get your TM.  Then use this TM to pre-translate the source and translate whatever's missing.

Studio 2021 has a filetype for YAML, and if you have an earlier version you can use this from the appstore:

https://appstore.sdl.com/language/app/file-type-definition-for-yaml/778/

If this isn't the usecase then you'll need to explain what it is?

Paul

In a way this is a very similar question to this one: https://community.sdl.com/product-groups/translationproductivity/f/regex_and_xpath/32238/custom-bilingual-xml-with-html-embedded-content/106300#106300

As demonstrated, I found a way to work with bilingual web files, but it is not a very pretty workflow.

I think it would be a real asset if this was possible natively in Studio, surely with xml, but maybe also with yaml. In web contexts, you often have an initual translation, then both (all) languages get independently modified outside the TMS/CAT context for SEO purposes etc. Then the source language is updated and this has to be reflected in the target, but without losing all optimization that has happened since. For us, this is the classic case where we need to work with a bilingual file.

Bilingual Excel is sometimes okay, but has quite a few limitations, e.g. no segmentation, no real handling of HTML entities ... It's great, a Swiss army knife for translation purposes. But if you are a butcher or a chef and you work with a knife all day every day, you are not going to choose a Swiss army knife...

Daniel

There is already a filetype for YAML on the appstore that can handle bilingual YAML, in fact it can handle multilingual YAML but only one pair at a time:

https://appstore.sdl.com/language/app/file-type-definition-for-yaml/778/

The PO, YAML, JSON filetypes from Remy Blaettler (Supertext) are a lot more versatile and useful than the out of the box filetypes.

However, I don't think they'll be helpful for the case in this thread which seems to be more of an alignment problem given the user has multiple monolingual files, not bilingual files.

Paul

Paul said:

this thread which seems to be more of an alignment problem given the user has multiple monolingual files, not bilingual files.

My workflow (and I would think any workflow that segments a bilingual source, however many files it is in) does the alignment as a necessary part of the workflow, and I would hope for that to be possible without leaving the Studio cosmos, so to speak. I think you have all the components, you would just need to link them together.

It's roughly like this: Read both source and target as monolingual content from however many files, initial segmentation by elements, fields or whatever the fixed unit is, then attempt segmentation within and align if possible.

Daniel

Paul

Paul said:

this thread which seems to be more of an alignment problem given the user has multiple monolingual files, not bilingual files.

My workflow (and I would think any workflow that segments a bilingual source, however many files it is in) does the alignment as a necessary part of the workflow, and I would hope for that to be possible without leaving the Studio cosmos, so to speak. I think you have all the components, you would just need to link them together.

It's roughly like this: Read both source and target as monolingual content from however many files, initial segmentation by elements, fields or whatever the fixed unit is, then attempt segmentation within and align if possible.

Daniel

Thank you. Maybe I'm just misunderstanding you, but my point is that in this thread the OP said this:

Laura H said:

my client has a database in Locale App, from where it's possible to export databases but only in monolingual format.

So she has monolingual source files, and she needs to translate a monolingual source file.  So where does your workflow fit in this scenario?  Apologies if I'm just being dumb here.

Paul

For my workflow it does not matter whether source and target are stored in one file or in two "parallel" files. The goal is to produce an XLF file with source and target so that the translator can work with the existing translations as a starting point.

ok - I see where you're coming from now.  I was being dumb!

Can't you align and then save as an sdlxliff?  I think the alignment is important because if it doesn't match exactly you have an opportunity to change it.  But if it does it's only a couple of clicks and you're done.

Paul

Paul said:

Can't you align and then save as an sdlxliff?

Of course you can, and in this case it may be the quickest way. But I often have projects with tens or hundreds of small files. I need a way to batch process this. My workflow is far from ideal, I think because I cobbled it together from freely available tools. If this was properly done within the framework of Studio, it might be a smooth, fast, easy to use process. Alas, I am not a developer.

Daniel

I'd be interested to see what your process is.  Can you record a small demo perhaps and share it?  Maybe something could be done once we see what you're doing.  I don't think this is a common thing... although happy to be corrected on this... but based on this it probably won't be something the core product will be interested to look at, but perhaps it's something we could consider improving with the community developers.

Exactly what I'm trying to achieve but my yaml source files seem to be a "collection" of translated units, rather than a complete set of units (whether translated or not). So I'm left with a mismatched alignment since the translations are only partially done so far.